[None][feat] Add NCCL device kernels for AR+RMS fusion

.codespellignore

@@ -0,0 +1 @@
+commIter

.pre-commit-config.yaml

@@ -1442,7 +1442,7 @@ repos:
         additional_dependencies:
         - tomli
         # add ignore words list
-        args: ["-L", "Mor,ans,thirdparty", "--skip", "ATTRIBUTIONS-*.md,*.svg", "--skip", "security_scanning/*"]
+        args: ["-L", "Mor,ans,thirdparty", "--skip", "ATTRIBUTIONS-*.md,*.svg", "--skip", "security_scanning/*", "-I", ".codespellignore"]
 -   repo: https://github.com/astral-sh/ruff-pre-commit
     rev: v0.9.4
     hooks:

cpp/tensorrt_llm/kernels/CMakeLists.txt

@@ -86,3 +86,4 @@ add_subdirectory(groupRmsNormKernels)
 add_subdirectory(llama4MinLatencyKernels)
 add_subdirectory(dsv3MinLatencyKernels)
 add_subdirectory(causalConv1d)
+add_subdirectory(nccl_device)

cpp/tensorrt_llm/kernels/customAllReduceKernels.h

@@ -58,6 +58,7 @@ enum class AllReduceStrategyType : int8_t
     LOWPRECISION = 6,
     MNNVL = 7,
     NCCL_SYMMETRIC = 8,
+    NCCL_DEVICE = 9,
 };
 
 enum class AllReduceStrategyConfig : int8_t

cpp/tensorrt_llm/kernels/nccl_device/CMakeLists.txt

@@ -0,0 +1,39 @@
+#
+# SPDX-FileCopyrightText: Copyright (c) 1993-2025 NVIDIA CORPORATION &
+# AFFILIATES. All rights reserved. SPDX-License-Identifier: Apache-2.0
+#
+# Licensed under the Apache License, Version 2.0 (the "License"); you may not
+# use this file except in compliance with the License. You may obtain a copy of
+# the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
+# WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
+# License for the specific language governing permissions and limitations under
+# the License.
+#
+
+# CMakeLists.txt for nccl_device This directory contains CUDA kernels and host
+# launcher code
+
+# Enable CUDA
+enable_language(CUDA)
+
+# Create CUDA library
+add_library(tensorrt_llm_nccl_device config.cu)
+
+# Set properties for the CUDA library
+set_target_properties(
+  tensorrt_llm_nccl_device
+  PROPERTIES CUDA_STANDARD 17 CUDA_SEPARABLE_COMPILATION ON
+             POSITION_INDEPENDENT_CODE ON)
+
+# Include directories
+target_include_directories(
+  tensorrt_llm_nccl_device PUBLIC ${CMAKE_CURRENT_SOURCE_DIR}
+                                  ${CMAKE_CURRENT_SOURCE_DIR}/../..)
+
+# Link libraries
+target_link_libraries(tensorrt_llm_nccl_device tensorrt_llm_common)